In [1]:
import pandas as pd
import numpy as np

import geopandas as gpd
import json

import statsmodels

import plotly
import plotly.express as px
import plotly.io as pio
# pio.renderers.default='jupyter'
# pio.renderers.default='jupyterlab'

# allow full interactivity offlice
plotly.offline.init_notebook_mode()

Load Data¶

Giffords Gun Law Scorecard¶

Load the Giffords Gun Law Scorecard for each state. Note, original scores have been coerced into a 5-point Likert scale and a numeric grade has also been assigned to each letter grades.

In [2]:
scorecard_df = pd.read_csv('giffords_gun_law_scorecard.csv')
scorecard_df.head(n=10)
Out[2]:
state_abbrev giffords_grade grade
0 NY A 4.0
1 NJ A 4.0
2 MD A 4.0
3 MA A 4.0
4 IL A 4.0
5 HI A 4.0
6 CT A 4.0
7 CA A 4.0
8 WA B 3.0
9 VA B 3.0

Firearm Related Deaths¶

Load the firearm related mortality data for each state from the CDC. Data was acquired using the Socrate API and wrangled for this presentation. See DataWrangling.ipynb for details.

In [3]:
firearm_related_deaths_df = pd.read_csv('tidy-489q-934x-firearm-related-injury.csv')
firearm_related_deaths_df.tail(n=10)
Out[3]:
year_and_quarter state state_abbrev mortality_per_100k
540 2022 Q3 South Dakota SD 16.7
541 2022 Q3 Tennessee TN 21.6
542 2022 Q3 Texas TX 16.1
543 2022 Q3 Utah UT 12.4
544 2022 Q3 Vermont VT 13.4
545 2022 Q3 Virginia VA 15.3
546 2022 Q3 Washington WA 13.0
547 2022 Q3 West Virginia WV 17.9
548 2022 Q3 Wisconsin WI 14.5
549 2022 Q3 Wyoming WY 23.6

Make state, state abbreviation, and Giffords grade columns categorical.

In [4]:
firearm_related_deaths_df = firearm_related_deaths_df.astype({'state': 'category', 'state_abbrev': 'category'})
print(firearm_related_deaths_df.dtypes)

scorecard_df = scorecard_df.astype({'state_abbrev': 'category', 'giffords_grade' : 'category'})
print(scorecard_df.dtypes)

year_and_quarter        object
state                 category
state_abbrev          category
mortality_per_100k     float64
dtype: object
state_abbrev      category
giffords_grade    category
grade              float64
dtype: object

Retain only the latest data for firearm related deaths.

In [5]:
deaths_q3_2022 = firearm_related_deaths_df[firearm_related_deaths_df['year_and_quarter'] == '2022 Q3']
deaths_q3_2022.head()
Out[5]:
year_and_quarter state state_abbrev mortality_per_100k
500 2022 Q3 Alabama AL 25.6
501 2022 Q3 Alaska AK 23.3
502 2022 Q3 Arizona AZ 20.9
503 2022 Q3 Arkansas AR 22.4
504 2022 Q3 California CA 9.1

Merge firearm related deaths and Giffords scrore datasets.

In [6]:
df = deaths_q3_2022.merge(scorecard_df, on='state_abbrev')
df = df.sort_values(by=['giffords_grade'], ascending=True)
df = df.reset_index(drop=True)
df.head()
Out[6]:
year_and_quarter state state_abbrev mortality_per_100k giffords_grade grade
0 2022 Q3 Illinois IL 14.7 A 4.0
1 2022 Q3 Maryland MD 13.5 A 4.0
2 2022 Q3 New Jersey NJ 5.3 A 4.0
3 2022 Q3 Hawaii HI 4.3 A 4.0
4 2022 Q3 Massachusetts MA 3.9 A 4.0

Choropleth map of Firearm Mortality¶

Load GeoJSON data that was previously downloaded from here.

In [7]:
us_states = json.load(open('states.geojson', 'r'))

Add a new property 'id' to features - 'id' is the default name of column that is used to map values from the dataset ('state') to the corresponding State in GerJSON data.

In [8]:
for feat in us_states['features']:
    feat['id'] = feat['properties']['NAME']

Display Firearm Mortality by State on map.

In [9]:
fig = px.choropleth(df, locations='state', scope='usa', geojson=us_states,
                    color='mortality_per_100k', 
                    hover_name='state', 
                    hover_data=['giffords_grade'],
                    color_continuous_scale='Portland',
                    color_continuous_midpoint=np.average(df['mortality_per_100k']),                
                    labels={
                     "mortality_per_100k": "Firearm Mortality per 100,000 persons",
                     "giffords_grade" : "Giffords Gun Law Scorecard",
                     "state" : "State"
                    },
                    title='Firearm Mortality and Gun Law Scorecard by State (Hover over map for details)')


fig.show()

Heatmap (Treemap) of Firearm Mortality grouped by Giffords Gun Law Scorecard¶

The Heatmap makes reveals the following trends:

  • States with the strictest gun laws have the lowest firearm related mortality rates.
  • Outliers: e.g. New Mexico, which has extremely high firearm related mortality despite a non-failing gun law score. Other outliers include Utah, Maine, Iowa, and New Hampshire. Explaining outliers is not in scope for this presentation.
In [10]:
import numpy as np
fig = px.treemap(df, path=[px.Constant('Giffords Gun Law Scorecard'), 'giffords_grade', 'state_abbrev'], values='mortality_per_100k',
                 color='mortality_per_100k', 
                #  labels={"mortality_per_100k": "Firearm Mortality per 100,000 persons"},
                 labels={
                     "mortality_per_100k": "Firearm Mortality per 100,000 persons",
                     "state" : "State",
                     "grade" : 'Giffords Gun Law Scorecard',                     
                     "giffords_grade" : 'Giffords Gun Law Scorecard'
                 },

                 height=600,
                 width=1000,
                 title='Trend: States with <b>Stronger</b> Firearm Control Laws have <b>Lower</b> Firearm Mortality<br>(Hover over treemap for details)</br>',
                #hover_name='label',
                #  hover_data=['mortality_per_100k'],
                 color_continuous_scale='Portland',
                 color_continuous_midpoint=np.average(df['mortality_per_100k'])
                )

fig.show()

Scatterplot with trend line using Ordinary Least Squares¶

In [11]:
fig = px.scatter(df, x="grade", y="mortality_per_100k",
                 color="mortality_per_100k", 
                 labels={
                     "mortality_per_100k": "Firearm Mortality per 100,000 persons",
                     "state" : "State",
                     "grade" : 'Giffords Gun Law Scorecard',                     
                     "giffords_grade" : 'Giffords Gun Law Scorecard'
                 },
                 hover_name='state', 
                 hover_data=['giffords_grade', 'mortality_per_100k'], 

                 color_continuous_scale=px.colors.diverging.Portland,
                 color_continuous_midpoint=15,
                 
                 height=600,
                 width=1000,

                 trendline='ols',
                 title='Trend: States with <b>Stronger</b> Firearm Control Laws have <b>Lower</b> Firearm Mortality<br>(Hover over points for details)</br>'
            )

fig.update_xaxes(type='category')
fig.update_xaxes(tickmode='array', tickvals=df['grade'], ticktext=df['giffords_grade'])
fig.update_layout(
          xaxis_title='Giffords Gun Law Scorecard for States',
          yaxis_title="Mortality per 100,000 persons"
)

#fig.update_traces(visible=False, selector=dict(mode="markers"))
fig.show()

Heatmap showing Gun Mortality by State and Gun Law Grade¶

In [12]:
import plotly.graph_objects as go
import datetime
import numpy as np
np.random.seed(1)

states = list(df['state_abbrev'])
grades = list(df['giffords_grade'])
mortality = list(df['mortality_per_100k'])

base = datetime.datetime.today()
dates = base - np.arange(180) * datetime.timedelta(days=1)

fig = go.Figure(data=go.Heatmap(
        z=mortality,
        x=states,
        y=grades,
        # hoverinfo=['x', 'z'],
        hovertemplate='State: %{x}<br>Grade: %{y}<br>Firearm Mortality: %{z}<extra></extra>',
        colorscale='Portland'))

fig.update_layout(
    title='States with the <b>Stronger</b> Gun Laws have the <b>Lower</b> Firearm Mortality<br>(Hover over heatmap for details)',
    xaxis_nticks=36)

fig.update_layout(
    xaxis_title="State",
    yaxis_title="Giffords Gun Law Scorecard",
    legend_title={'text': "Firearm Mortality per 100,000 Persons"}
)

fig.layout.height = 500
fig.layout.width = 800

fig.show()
In [ ]: